| Sign In to gain access to subscriptions and/or personal tools. |
Does Layering Minimize Shrinkage Stresses in Composite Restorations?
1 (PCT 117) Department of Preventive and Community Dentistry and Pedodontology, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands; Correspondence: *corresponding author, R.Kuijs{at}dent.umcn.nl
Polymerization shrinkage of resin composites may impair restoration longevity. It is hypothesized that layering, rather than bulk, techniques result in less stress in the tooth-restoration complex. The aim of this study was to compare shrinkage stresses for different restorative techniques used for cusp-replacing restorations with direct resin composite. In a 3-D FE model, the dynamic process of shrinkage during polymerization was simulated. Time-dependent parameters (shrinkage, apparent viscosity, Young’s modulus, Poisson ratio, and resulting creep), which change during the polymerization process, were implemented. Six different restorative procedures were simulated: a chemically cured bulk technique, a light-cured bulk technique, and 4 light-cured layering techniques. When polymerization shrinkage is considered, a chemically cured composite shows the least resulting stress. The differences seen among various layering build-up techniques were smaller than expected. The results indicate that the stress-bearing locations are the interface and the cervical part of the remaining cusp.
Key Words: FE model • polymerization stress • composite • cusp replacement
The setting of composite restorative materials is accompanied by shrinkage due to the formation of polymers out of monomers. By the use of specific restorative techniques, stresses resulting from constrained shrinkage might be reduced. However, it is not clear which adhesive restorative technique should be used to reduce the shrinkage stresses. Applying the composite in layers instead of using a bulk technique is suggested to reduce the shrinkage stresses (Donly and Jensen, 1986) and to reduce microleakage at the interface (Tjan et al., 1992). Another study reported that bulk techniques resulted in lower stresses (Versluis et al., 1997). It has also been reported that there is no difference between bulk or layering techniques when pain or microleakage is concerned (Opdam et al., 1998a,b). The shrinkage stresses may also be influenced by the composition of the restorative material (Donly et al., 1987; Suliman et al., 1994; Swift et al., 1996). Polymerization of a chemically cured composite takes more time, which results in lower polymerization stresses than when a light-cured composite is used. Slow-start polymerization may reduce the shrinkage stresses by prolonging the curing time (Feilzer et al., 1995; Yoshikawa et al., 2001). Many variables seem to influence polymerization stresses. With in vitro load tests and leakage tests, different aspects of polymerization shrinkage have been studied, but these tests do not show the location of the stresses in the tooth-restoration complex. It is important to locate a high-stress area because that is most likely the place where failure of the restoration initiates. A popular exploratory laboratory technique is Finite Element (FE) model analysis, consisting of computer simulation of mechanical processes. With an FE analysis, it is possible to visualize internal stresses, and on that basis predictions can be made about failure and its location (Versluis et al., 1996; Proos et al., 2002; Barink et al., 2003). FE models can be useful for our understanding of the physical process of polymerization, shrinkage, and resulting stresses. When the polymerization process is modeled in an FE model, this is generally done by using the thermal expansion coefficient as a substitute for shrinkage (Versluis et al., 1996; Winkler et al., 1996). During the polymerization process, many factors, such as viscosity and Young’s modulus, continuously change and will therefore influence the resulting stress. The polymerization process has been modeled in a simplified 2-D model (Hübsch et al., 2000). Although their axi-symmetric FE model was adequate for a Class I restoration, it cannot be applied to more complex geometric situations, such as MOD or cusp-replacing restorations. The aim of our study was to compare shrinkage stresses of different restorative techniques used in a large cusp-replacing composite restoration to see which technique results in the lowest polymerization stresses. The hypothesis was that the chemically cured restoration will develop the lowest stresses and that bulk light-cured restorations will show higher polymerization stresses than restorations built up in layers. We developed a highly detailed 3-D FE model of a premolar to incorporate the complex geometric effects of different layering techniques. The dynamic process of stress development during the curing process of the composite was modeled. Special attention was paid to the developed stresses at the interface for the different restorative methods.
The FE model used for this study is based upon an earlier FE model (Verdonschot et al., 2001; Barink et al., 2003). Micro-CT scans were used to model the inner geometry of dentin, pulp, enamel, and a cusp-replacing restoration in an upper premolar. The outline of the preparation ended in enamel, with a butt joint cervical margin in the boxes and the buccal outline ending in a 45° angle (Fig. 1  ). The FE model used for this study consists of 7797 iso-parametric elements and 9500 nodes. All materials were assumed to be linear elastic (except for the composite material) and isotropic. The composite material was assumed to be rigidly fixed to the tooth tissue by means of ties, which prescribed equal displacement of both the composite and the tooth tissue-node. The interface stresses were calculated from the internal nodal forces at the interface. Using local coordinate systems and actual contact surface at the interface, we transformed these nodal forces into interface stresses. This method ensured compatibility of the interface stresses, which is not achieved when interface stresses are calculated by extrapolation of element integration point values.
The model included the main time-dependent composite polymerization parameters: shrinkage, Young’s modulus, Poisson ratio, and the resulting creep. These parameters were adapted during the simulations with user-defined software subroutines. This means that the changes in shrinkage, Young’s modulus, Poisson ratio, and creep that take place during polymerization of each layer of the composite were implemented in the model. The parameters were calculated from the measured data for a chemically cured composite (Feilzer et al., 1990; Barink et al., 2003). The shrinkage was modeled by means of temperature-dependent expansion. Decreasing temperature and the use of a non-physiological expansion coefficient resulted in the necessary shrinkage values. The time-dependent Poisson ratio is interpolated based on the fact that polymers below the rubber point are (almost) incompressible and that the Poisson ratio of a solid composite has been measured to be 0.21 (Akinmade and Nicholson, 1995). Creep was modeled with an associative creep law (Hübsch, 1995) based on time-dependent apparent viscosity (Hübsch et al., 2000). Validation of the algorithms used in our model was described in a previous publication (Barink et al., 2003).
Two restorative approaches were simulated in the model (Fig. 1 To model the different layering techniques, we divided the elements representing the restoration into several element sets. Each set represented a layer of the filling procedure. By choosing a different order of layers, we could simulate a different restorative approach. Each layer was polymerized separately, with a maximum thickness of 2 mm. For the bulk chemically cured composite, a setting time of 15 min was modeled. For the bulk light-cured composite, this time was re-scaled to 30 sec; hence, all relevant parameters were also re-scaled. For each layer of the other procedures, a setting time of 30 sec was modeled, with a pause of 1 min between layers. After the last layer was set, the stresses were calculated. Changes in temperature during polymerization were assumed to be minimal and therefore were not taken into account. For all light-cured restorations, it was assumed that the degree of conversion was the same as for the chemically cured material. We used FE software (MARC/McNeal Schwendler, Palo Alto, CA, USA) to calculate the principal stresses and principal strains separately for the different materials (enamel, dentin, and composite). At the interface, normal and shear stresses between restoration and tooth tissue were calculated.
In general, the locations of the high stress concentrations resulting from the polymeriza-tion shrinkage were similar for all restorative procedures. A representative stress distribution is shown in Fig. 2 . The highest tensile stresses were located in the enamel at the cervical part of the non-restored cusp (Fig. 2B). High tensile stresses were also seen at the restoration periphery (Figs. 2A, 2B) and at the upper 2 mm of the cavosurface margin of the remaining cusp (Fig. 2A). High interfacial shear stresses were located at the restoration outline and at the internal line angles of the preparations (Fig. 2C).
Fig. 3 shows the principal shear stresses with their corresponding volume percentages in the enamel, dentin, and composite. In dentin, about 85% of the volume experienced stresses lower than 2 MPa (Fig. 3B), while for enamel this volume is about 15% (Fig. 3A). The highest stresses were seen in the enamel. Approximately 3% of the volume experienced a shear stress higher than 8 MPa (Fig. 3A).
When the restoration procedures are compared, the shear stresses were lowest for chemically cured composite. This was best demonstrated in the lower-stress regions of the composite material (Fig. 3C  ). In the higher-stress regions (higher than 6 MPa), the differences between the chemically cured and the light-cured procedures were small. For all procedures, 1-2% of the interface surface between tooth and restoration experienced a maximum shear stress of 12 MPa or more. For the major part (approximately 80%), the interface was subjected to stresses lower than 2 MPa (Fig. 4).
Stress in a composite restoration is considered a negative characteristic. Polymerization shrinkage can result in tensile and shear stresses. Therefore, techniques are applied to minimize the stresses due to the polymerization shrinkage. In this study, a 3-D FE model showed resulting internal stresses. It was also possible to locate highly stressed regions. This feature is an advantage of 3-D FE models over other established research techniques.
From studying all data, we concluded that the restoration-tooth complex is most likely to fail at the interface rather than in the composite or tooth material (Fig. 2 For the models in this study, an adhesive layer was modeled that can be described as a robust interface between the cavity surface and the restoration which can resist extreme high stress before rupture. The interface was modeled as an adhesive layer 0 µm thick, while in some other FE models the adhesive layer was 50 µm thick (Ausiello et al., 2002). In the latter study, it appeared that a thicker adhesive layer reduces stress in the tooth-restoration complex. Incorporating a thick interface might have reduced the small differences between the restoration approaches, as found in this study. Since in vitro study indicates that the thickness of an adhesive layer may vary between 10 and 60 µm (Kuijs et al., 1998), it is not clear which number to choose in this respect. Above all, the present FE study was aimed to explore stress distribution in the restorations and at the interfaces with different layering sequences of the restorations. An elastic interface might then be a confounder. In the lower-stress regions, differences were seen between the chemically cured and the light-cured restorations, while there were only minor differences between the different light-cured restoration techniques. This suggests that the polymerization time of the composite material is more important than the restorative procedure in preventing stresses. This supports the advantage claimed in the slow-start polymerization theory (Feilzer et al., 1995; Versluis et al., 1998).
In two of the layering procedures, the center of the cavity was restored first, while in the other two layering procedures the cusp was restored first, and an MOD-like cavity remained to be filled (Fig. 1
Failure probability of the materials and the interface is often estimated by comparison of the peak stress values with strength values. A simulation of the polymerization process of chemically cured resin composite revealed that failure of the interface is much more probable than failure of the composite material (Barink et al., 2003). A literature analysis of shear bond strengths showed an average shear bond strength of 12 MPa between dentin and composite resin (al-Salehi and Burke, 1997). If shear stresses higher than 12 MPa are considered in the FE model, 1-2% of the interface might be debonded (Fig. 4 The stresses in this model were pure polymerization stresses calculated during and directly after the restorative procedure. The percentages of debonded surfaces can increase when load is applied to the restoration. The consequences of this debonding are not considered in this model. Additionally, it can be expected that the stresses decrease during the first hours after polymerization (Barink et al., 2003). For this reason, it can be advisable to instruct the patient not to load the restoration for a certain period. Based on the present results, it can be concluded that when polymerization shrinkage is considered, a chemically cured composite shows the least resulting stress. The differences seen between various layering build-up techniques were smaller than expected. For a cusp-replacing restoration, the order in which the composite layers are placed has almost no influence on the developed stress. Therefore, specific application methods concerning the order of layers to reduce polymerization stress are not indicated. The stress-bearing locations are the interface between the preparation and the restoration and at the cervical part of the remaining cusp.
This study was supported by the University of Nijmegen and is part of the research program "Oral diseases and musculoskeletal disorders" of the Faculty of Medical Sciences of the University of Nijmegen. The authors thank Paul van der Mark for all his work on the project. Received for publication February 11, 2003. Revision received August 29, 2003. Accepted for publication September 8, 2003.
Journal of Dental Research, Vol. 82, No. 12,
967-971 (2003) This article has been cited by other articles:
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
TOP
ABSTRACT
INTRODUCTION
